From Language Log: The snow words myth: progress at last [1]

It is not all gloom as regards the media's treatment of language. There are happy stories too. Ash Asudeh just sent me a little "whaddya know" piece headed "Snow Speak" that he scanned from an airline magazine (Holland Herald, published by KLM). It had an illustrative drawing of an Arctic hunter, and it was about snow words. Yawn, I thought. But this one was a real surprise. They had actually been talking to a linguist, it seems, or had at least once met one in a bar somewhere, and although what they said was not accurate, it was a lot closer to being accurate than the familiar nonsense that has been repeated so many times:

The idea that Inuit people have many more words for snow than English speakers is a myth. Most Inuit languages are "polysynthetic". Whereas English uses separate words in the sentence "the snow under the tree" an Inuit person would express this in one word. In fact, English has more words for different types of snow than most Inuit languages.

This still hasn't got everything right. An unsympathetic judgment would be that it's stuffed full of mistakes: (1) the language family is generally called Eskimo or Eskimoan, because it includes the Yup'ik languages of Siberia and Alaska as well as the Inuit languages from the northeastern half of Alaska across Canada to Greenland; (2) all eight Eskimoan languages are polysynthetic to a high degree, not just most; (3) the distinction between bases and derived words isn't even hinted at here, but it's crucial; (4) "the snow under the tree" is not a sentence, it's a noun phrase; (5) I don't think the definite articles in the latter phrase would typically come across in the meaning of a derived word, so the example is a bad one; (6) the point is not about what an Inuit person would do, it's about the structural resources an Eskimo language provides; (7) it's not clear that English has more words (who's counting?), it's just that it appears to be roughly comparable by most sensible ways of counting distinct genuinely snow-related lexeme roots. The point is that we want to count one for each family of derived words like snow, snowy, snowing, snowlike, snowstorm, etc.; if you don't do that, then Eskimoan languages not only have millions of words for snow, they have millions of words for fish, millions of words for coffee, millions of words for absolutely anything, which makes the whole discussion
irrelevant to anything about snow.